Urban Thermal Anisotropy: A Comparison Among Observational and Modeling Approaches at Different Time Scales

The directional variation in upwelling thermal radiance (referred to as "thermal anisotropy") is one of the important issues in retrieving representative urban land surface temperature (LST) from remote sensing data. Yet, there is a gap between urban thermal anisotropy assessed from different observational and modeling approaches at different time scales. Thermal anisotropy at the satellite observing scale (~1 km) is usually derived from multiangular thermal observations that are a result of temporal aggregation over long time series. How well this type of anisotropy estimate represents the instantaneous thermal anisotropy (e.g., from airborne and some modeling techniques) at the satellite scale remains unclear. To close this knowledge gap, we comprehensively compare Moderate-Resolution Imaging Spectroradiometer (MODIS)-derived anisotropy and quasi-simultaneous airborne observations in urban areas and assess the factors that control anisotropic features of LST at seasonal and diurnal scales. Furthermore, we use model simulations to separately assess the impacts of 1) weather variability on anisotropy and 2) solar angle variation on zenithal variation in anisotropy, both of which have been largely simplified in the MODIS-derived approach. The instantaneous model-derived or airborne-measured anisotropy in early morning (~10:00) and late afternoon (~14:00) is closer to the seasonally aggregated MODIS-derived anisotropy from Terra-Day (with a median overpass time of 11:00) and Aqua-Day (13:00), respectively, whereas it has a good representation during winter. The investigation of the diurnal and seasonal characteristic of MODIS-derived anisotropy is a critical step needed to understand its performance for global urban thermal anisotropy profiles.